Stochastic Logic Programs: Sampling, Inference and Applications

TL;DR

This paper introduces algorithms for inference in stochastic logic programs, demonstrates their use in representing priors in machine learning, and applies Markov chain sampling methods for posterior inference.

Contribution

It presents novel inference algorithms for stochastic logic programs and explores their application in Bayesian modeling and sampling-based posterior inference.

Findings

Algorithms for exact and approximate inference are developed.

SLPs can effectively represent prior distributions in machine learning.

Metropolis-Hastings sampling is applied for posterior distribution sampling.

Abstract

Algorithms for exact and approximate inference in stochastic logic programs (SLPs) are presented, based respectively, on variable elimination and importance sampling. We then show how SLPs can be used to represent prior distributions for machine learning, using (i) logic programs and (ii) Bayes net structures as examples. Drawing on existing work in statistics, we apply the Metropolis-Hasting algorithm to construct a Markov chain which samples from the posterior distribution. A Prolog implementation for this is described. We also discuss the possibility of constructing explicit representations of the posterior.